A Study of the Parallelization of the Multi-Objective Metaheuristic MOEA/D

MOEA/D is a multi-objective metaheuristic which has shown a remarkable performance when solving hard optimization problems. In this paper, we propose a thread-based parallel version of MOEA/D designed to be executed on modern multi-core processors. Our interest is to study the potential benefits of the parallel approach in terms of speedups and the quality of the obtained Pareto front approximations when solving a benchmark composed of nine problems. The obtained results on two different multi-core based machines indicate that notable time reductions can be achieved. We have also found out that, with a few exceptions, there are not significant differences in terms of solution quality among the sequential MOEA/D and the parallel versions of it when using up to eight threads. Key words: Multi-Objective Optimization, Metaheuristics, Parallelism, Multi-core processors

• Type: Conference
• Year: 2010
• Where: LION
• Authors: Antonio J. Nebro, Juan J. Durillo
• Keywords: LION 2010, Modern Multi-core Processors, Multi-core Processors, Optimization, Parallel Versions